The very old represent the fastest growing segment of the elderly population, those with the most brittle antidepressant treatment response, and yet the least studied in randomized clinical trials. Hence, information on the optimal strategies for the long-term clinical management of such patients is greatly needed. In our recently completed studies of maintenance therapies in late-life depression (MTLD-1), we have learned that nortriptyline (NT) and interpersonal psychotherapy (IPT), either alone or in combination, are substantially better than placebo In preventing recurrences of geriatric major depression and that combined treatment works better than monotherapy in maintaining recovery. However, long-term treatment response in patients aged 70 and above appears much more variable and brittle than in 60-69 year-olds, with monotherapy performing less well than combined treatment with medication and psychotherapy. Despite increasing use of SSRI antidepressants in old-age depression, there are no controlled evaluations of their long-term efficacy. Thus, in MTLD-2, we propose to test the hypothesis that combined maintenance treatment with paroxetine (PX) and IPT will be superior to either alone and to placebo in maintaining recovery and in reducing long-term treatment-response variability in the 70+ year old patients. As well, cost-benefit analyses of combined versus monotherapeutic strategies are necessary and also will be provided by MTLD-2. Finally, in order to address the question: 'Which treatments work best for which patients?,' we will determine moderators of long-term treatment- response variability in geriatric depression, including cognitive impairment and brain structural changes. Two hundred subjects aged 70 and above with current major depression (non-psychotic, non-bipolar) will receive acute and continuation treatment with PX and IPT. Patients who recover (estimated n = 125) will receive maintenance treatment, with random assignment to one of four conditions: 1) medication clinic + PX; 2) medication clinic + placebo (PBO); 3) IPT + PX; and 4) IPT + placebo. Maintenance treatment will last two years or until recurrence of major depression, whichever occurs first. Rates of recurrence and time to recurrence in each condition will be contrasted via survival analysis. We will also examine the cost- effectiveness of combined versus monotherapy in maintenance treatment and determine moderating variables of long-term treatment response. This proposal is the competing renewal of MH43832.